You are on page 1of 3

the transmitter architecture for N-x-OFDMA system with two stages of Noise shaping process is shown in fig.2.

Let us consider = 0, 1…N-1 be the input block of frequency domain modulated symbols. At here N is the
number of Symbols in each Component Carrier (CC). If clipping is performed with the Nyquist sampling rate in
the discrete time domain it may affect the high frequency components in the in-band and the clipping noise
directly sits in the desired in-band which may degrade the BER performance. An oversampling rate factor of L ≥
4 will be used so that the PAR before D/A conversion accurately describes the one after D/A. To examine the
out-band distortions introduced by the noise shaping algorithm, Oversampling is also necessary.

--- represents the oversampled discrete-time domain OFDM symbols . it can be calculated as follows

Fig. 3 shows the Block diagram of the First stage Noise shaping process . either Cartesian or polar clipping can
be used To clip the peak of teh signal, the In-phase and qudrrature components are clipped independently with
the Cartesian clipping.the phase of signal will be preserved and the magnitude will be clipped by using polar
clipping.with polar clipping. At here we using the polar clipping to limit the PAPR of the signal. the polar
clipping provides better results in terms of overall signal distortion (i.e. lower EVM and ACLR performance).
the PAPR of the signal can be maintained at a desired level By selecting an appropriate threshold Amax. Polar
clipping of the time domain discrete data ------- with clipping threshold Amax results in the following equation.

To generate the clipping noise , polar clipped signal is subtracted from the delayed version of the incoming
OFDM symbols

the noise shaping low pass filter has the characteristics similar to that of a single carrier before it has been
unconverted. Raised-cosine function adopted the Spectrum shaping Low pass filter structure with roll-factor β
= 0.22.
At here h(n) represents impulse response of the Raised cosine filter.It is given by
At last a signal with lower PAPR is created by substracting spectrally shaped clipping noise from the delayed
version of the incoming OFDM signal.

The above technique doesn’t involves any hard clipping on the OFDM signal. So there will be minimal impact
on the in-band and out-band performance .After that individual component carriers are frequency shifted to
the band of interest in the given spectrum. Time domain data of each Component Carrier can be described as y
(i, j), here i is the ith Component carrier and j is the sub-frame.
Then the PAPR of aggregated k component carriers can be represented as

since few subcarriers of CC’s may add up constructively in time domain after Carrier aggregation ,So PAPR of
the aggregated carriers may increase even after clipping individual component carriers using Noise shaping
technique. We amy face this effect if the component carriers are Contiguous. To achieve the desired PAPR
further iterations of Noise shaping technique are required.

The clipping is performed on sufficiently oversampled (R ≥ 4) pass band carrier aggregated OFDM signals By
the second iteration of noise shaping process .then Oversampling results in lower in-band distortion upon
clipping the pass band signal because the clipping noise will transit into out-band which could be further
filtered out.
The Clipping noise is generated by subtracting the polar clipped signal from the delayed version of the
Composite pass band signal. It is further spectrally shaped by using noise shaping filter. filtering has
to be done for multiple frequency components in the spectrum to support multiple CC’s .The
spectrum of the complex composite signal might be asymmetrical . Then the number of taps required
to implement such filter is often quite high. Fig. 4.illustrates the process , the pass band clipping noise
will be down converted to baseband and perform low pass filtering ,then up convert the filtered
noise back to its carrier frequency. This approach provides more flexible multi-band filtering capability
that supports dynamically changing carrier configurations.

........ represents the inputs to the noise shaping filter. the down converted output noise with k component
carriers is given by




by raising cosine function The down converted clipping noise will be spectrally shaped . In next
step we do frequency shift of each filtered noise
component from baseband to corresponding center frequency. ........ represents the inputs to the noise shaping
filter. then the up converted noise output with k component carriers is given by

By frequency translation the individual k component carriers are summed to form composite pass band
clipping noise.


By noise shaping, the spectrally confined clipping noise is subtracted from an appropriately delayed version of
the CFR input signal which results in desired PAPR.
We simulates different scenarios using contiguous & non-contiguous frequency bands for carrier aggregation.
PAPR performance with and without Noise shaping technique is evaluated using the below equation.



PAPR performance can be compared by using Complementary cumulative distribution function (CCDF),It
shows the probability of the signal PAPR > PAPR0.







simulation of different scenarios are described inThis section. At here we considered 5 component carries, of
each component carrier has bandwidth of 5 MHz ,2560 subcarriers and each component carrier input 512, 16
QAM symbols.. So the total aggregated system bandwidth of 25 MHz Raised cosine filter with
β =0.22 is used for Noise Shaping.



Fig. 5 illustartes the comparison in PAPR performance between original signal, with only a single stage
Noise shaping process and the two stage Noise shaping process using QAM modulation of contiguous 5
CC’s aggregated

Fig. 6 illustartes the comparison in PAPR performance between original, single stage and two stages Noise
shaping process using QAM modulation of non-contiguous 2 CC’s aggregated (1st and 4th component carriers are
considered in simulation).

The approach will be similar to determine the PAPR performance of inter band non-contiguous carrier
aggregated OFDM signals.

Fig. 7 illustartes the power spectral density of original signal and the clipped signal of 5 CC’s aggregated. It
could be seen that the out of band spectral emissions are lower using the proposed two stages of Noise shaping
process.

Fig. 8 illustartes the power spectral density of original signal and the clipped signal of 2 Noncontiguous CC’s
aggregated. It could be seen that the Out of band spectral emissions are lower using the proposed two stages of
Noise shaping process.

1st and 4th component carriers of 5MHz bandwidth each are considered in simulation.


At here we have proposed a two stage Noise shaping process to reduce the PAPR while satisfying EVM and
ACLR /spectral mask constraints. For contiguous bands scenario the first stage of Noise shaping process
reduces the PAPR of the individual component carriers to some optimized threshold but the effect will be
minimal since after carrier aggregation some of the subcarriers of different CC’s may add up constructively
resulting in little improvement in PAPR. Then the second Noise shaping process is utilized to further reduce
the PAPR to desired level.For non-contiguous bands scenario, since the Component carriers are far apart, the
first noise shaping process itself results in higher PAPR improvement even after carrier aggregation since the
impact of component carriers adding up constructively in time domain will be minimal. The improvement in
PAPR with the use of two stages of Noise shaping process will be better in case of Non-contiguous scenario
compared to the contiguous.